Drones, digital twins, and deep learning for the infrastructure holding civilization up.
CAPTION: A drone hangs a few centimeters from a 500-foot tower, photographing a bolt no human would ever climb to reach. The reactor is live. The pilot is a neural net.
Someone, somewhere, is strapping into a harness right now to climb a 500-foot cell tower with a clipboard and a camera. It is dangerous, slow, and expensive - and PRENAV thinks it is faintly absurd. Why send a person up when a drone can fly the same steel, photograph every rivet, and hand the images to an AI that never blinks?
That is the whole premise. PRENAV digitizes critical infrastructure using drones, LiDAR, 3D analytics, and deep learning. Its web platform, PRENAV.XYZ, takes ordinary drone imagery and stitches it - via photogrammetry - into a detailed 3D digital twin of a structure. Then deep learning reads that twin like an X-ray, hunting for cracks, corrosion, and the kind of defects that turn into headlines if ignored.
The precision is the punchline: the system can resolve features smaller than 0.2mm on a surface - roughly twice the width of a human hair - from a camera hovering in the air. It has inspected bridges, dams, and towers. And, in a sentence that should not be possible, it completed the first-ever automated drone flights inside a functional nuclear reactor. No GPS. No pilot. Just a ground robot mapping the space and guiding the aircraft through it.
Approximate figures compiled from public filings and press coverage; some financials are estimates.
We use AI to make the world's critical infrastructure safer.
A drone - off-the-shelf or PRENAV's own, guided by a ground robot - flies the structure with centimeter accuracy, even where GPS fails.
Photogrammetry stitches the imagery into a detailed 3D digital twin you can inspect from a desk.
Deep learning scans the twin for cracks, corrosion, and defects down to 0.2mm.
PRENAV.XYZ delivers a change-tracked, actionable inspection - no rope, no scaffold.
A web-based platform for automated visual inspection of civil and industrial structures. Photogrammetry plus AI builds 3D digital twins, then deep learning finds defects finer than 0.2mm.
Procedurally generated, auto-labeled training images. Thousands of variations in geometry, texture, angle, and lighting - with bounding boxes and per-pixel labels, no manual annotation.
The original system: a ground robot scans and maps an environment, then guides a proprietary drone within centimeters of a structure - GPS-independent by design.
Stanford computer-music grad who ran product and marketing at EA and Lolapps, then managed robotics projects before founding PRENAV. The unlikely path from video games to nuclear reactors.
Part of the founding team that set out in 2013 to solve autonomous flight in complex, GPS-denied environments.
Co-founder helping assemble one of the top computer-vision and robotics teams, spanning aeronautics, embedded systems, and software.
The 2016 seed round was led by Crosslink Capital, with Haystack, Liquid 2 Ventures, and WI Harper Group. Public totals put cumulative funding in the range of $9M across seed rounds.
PRENAV starts in Redwood City with a goal: solve autonomous flight in complex environments.
Releases video of what it billed as the world's most precise commercial drone system; earns FAA Section 333 exemption for industrial inspections. Wins Robotic Film Festival and New Media Film Festival awards for precision light-painting.
Closes seed funding led by Crosslink Capital. Named to Bloomberg's "50 Most Promising Startups."
Announces partnership with the telecom giant for drone-based infrastructure inspection.
Named one of Forbes' "7 Industry-Shaking AI Startups" and featured among NVIDIA's GPU-powered startups at GTC.
Raises a further $1.5M to keep digitizing critical infrastructure.
The climber is gone. The tower still stands - and now it tells you, months in advance, exactly where it's going to fail.